Doubt Regarding the Deployment. about helmchart HOT 11 CLOSED

Oh boy, are you using Google Cloud by any chance? I had spent half a day troubleshooting when this happened to me. Because I'm a newbie, I think that at the end I ended up deleting and recreating the storage I had set up on the Cloud Console. I hope some of the more seasoned people in this forum can provide a better solution.

We do have the workaround of disabling persistent storage completely:
master.database.persistence=false and slave.database.persistence=false can be set, to disable persistenct completely. I just noticed that we don't have these two values on the README.md, though they do exist in values.yaml. Doing a change on the README.

from helmchart.

Hi @cakrit,
I don't use the Google Cloud. I am trying to deploy the most simple environment, with netdata collecting the physical resources from Kubernetes Master and Slaves.

I deactivate the persistent storage as you said, Thank you! The pods are started successfully but I don't see the metrics being collected. I cannot access the Master Netdata Dashboard. I attached the describe pods for the Master Pod.

Only for information, my environment has only one Master and one Worker, I installed the helm via the snap, and I am using the Kubernetes v1.14.

describeMaster.txt

from helmchart.

Check the notes shown after the installation to see how you'll be able to access the UI via the nginx proxy. They are not great, I've requested some help in #32.

Marking this as a bug with help wanted as well, for the original issue of the unbound volume claims.

from helmchart.

If I run kubectl get services. I have only two services, one for kubernetes and another one for netdata. I don't have the Nginx service. It's installed automatically via the helm netdata, or I need to configure it separated?

After the helm netdata start, the output brings the notes that you said and these resources description:
I suppose it's necessary the nginx proxy or the v1beta1/Ingress is my ingress point to access the netdata master dashboard?

RESOURCES:
==> v1/ClusterRole
NAME AGE
my-release-netdata 0s

==> v1/ClusterRoleBinding
NAME AGE
my-release-netdata 0s

==> v1/ConfigMap
NAME DATA AGE
netdata-conf-master 2 0s
netdata-conf-slave 5 0s

==> v1/DaemonSet
NAME DESIRED CURRENT READY UP-TO-DATE AVAILABLE NODE SELECTOR AGE
netdata-slave 2 2 0 2 0 0s

==> v1/Pod(related)
NAME READY STATUS RESTARTS AGE
netdata-master-0 0/1 ContainerCreating 0 0s
netdata-slave-5bs8z 0/1 ContainerCreating 0 0s
netdata-slave-n9pkc 0/1 ContainerCreating 0 0s

==> v1/Service
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
netdata ClusterIP 10.98.169.4 19999/TCP 0s

==> v1/ServiceAccount
NAME SECRETS AGE
netdata 1 0s

==> v1beta1/Ingress
NAME HOSTS ADDRESS PORTS AGE
netdata netdata.k8s.local 80 0s

==> v1beta2/StatefulSet
NAME READY AGE
netdata-master 0/1 0s

from helmchart.

With minikube I was able to connect directly, but on Google cloud I just had to use the nginx ingress. The helm install command in the readme is supposed to install the nginx-ingress for you as well. Can you delete it, install it again and see the output helm? It looks like something didn't go well.

from helmchart.

Hi guys,
I tried to delete @cakrit, but the result is always the same.

NAME: my-release
LAST DEPLOYED: Fri Jun 21 16:47:30 2019
NAMESPACE: default
STATUS: DEPLOYED

RESOURCES:
==> v1/ClusterRole
NAME AGE
my-release-netdata 0s

==> v1/ClusterRoleBinding
NAME AGE
my-release-netdata 0s

==> v1/ConfigMap
NAME DATA AGE
netdata-conf-master 3 0s
netdata-conf-slave 5 0s

==> v1/DaemonSet
NAME DESIRED CURRENT READY UP-TO-DATE AVAILABLE NODE SELECTOR AGE
netdata-slave 2 2 0 2 0 0s

==> v1/Pod(related)
NAME READY STATUS RESTARTS AGE
netdata-master-0 0/1 ContainerCreating 0 0s
netdata-slave-nf9zj 0/1 ContainerCreating 0 0s
netdata-slave-q9nbl 0/1 ContainerCreating 0 0s

==> v1/Service
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
netdata ClusterIP 10.97.59.0 19999/TCP 0s

==> v1/ServiceAccount
NAME SECRETS AGE
netdata 1 0s

==> v1beta1/Ingress
NAME HOSTS ADDRESS PORTS AGE
netdata netdata.k8s.local 80 0s

==> v1beta2/StatefulSet
NAME READY AGE
netdata-master 0/1 0s

The nginx ingress doesn't appear. However, the Netdata master is collecting the metrics I tested using the rest api on the netdata master pod IP:19999. I don't know why the nginx ingress port is not working. I can't access the netdata master dashboard externally, only locally.

from helmchart.

Hm... do this please:
helm install stable/nginx-ingress

I'll read up on the dependencies and what's supposed to happen. Maybe it's a step that we implied and we shouldn't have...

from helmchart.

Gents, it looks like we need a requirements.yaml with the nginx-ingress in it: https://github.com/helm/helm/blob/master/docs/helm/helm_dependency.md

I'll see if I can get to it soon, it should be a very easy PR for anyone who wants to hep.

from helmchart.

FWIW I don't think most charts that have ingresses (e.g. sonarqube, lamp or concourse) set up an explicit dependency to nginx-ingress. Probably this is because most users expect to manage the ingress controller as a separate install instead of as a dependency to any given chart (since usually only one controller is deployed per cluster), and different users might want to use different ingress controllers (e.g. traefik)

That said I don't think it's a problem as long as it's easy to disable the requirement via a condition 🙂

from helmchart.

We will continue this in #32, because it's linked to the lack of proper documentation / instructions on how to access the master node.

from helmchart.

Can I see render templates helm template --name my-release ./netdata? I don't think netdata must depend on ingress-controller.
On the first. You can turn off ingress-controller .ingress.enabled=false and use NodePort .service.type=NodePort (https://kubernetes.io/docs/concepts/services-networking/service/#nodeport ) the Kubernetes control plane will allocate a port from a range specified by --service-node-port-range flag (default: 30000-32767)
On the second. If you have another kind of ingress-controller. You can switch ingress.annotations.kubernetes.io/ingress.class=nginx to another (https://kubernetes.github.io/ingress-nginx/user-guide/multiple-ingress/). It can be gce, nginx, nginx-plus, traeffic or etc. Ingress-controller give abstraction above L7 Balancer

from helmchart.